Responsiveness of 2.3.36

it’s a while I don’t use Logic.
My latest version was 2.3.26. I was quite happy with that.

I’m running some Async serial analyzers together with some custom extensions on them.
I’ve downloaded the 2.3.36 version and the responsiveness of Logic got a lot worse.

Sometimes when i press the stop button nothing happens for >10 seconds, sometimes it doesn’t stop at all. I had to kill the process several times.
The acquisition image (while it is acquiring) is stuttering all the time.

Did you change something in the Async serial extensions processing?
Right now I went back to 2.3.26 and it works pretty well.

Thanks for reporting this!

Could you provide some more information?
Capture details: sample rate, which channels are enabled, which recording mode you’re using, how long you’re recording, and if you are using the glitch filter.

Thanks for the detail about the stop button delay. That indicates a very specific failure mode.

Could you also let us know if you’re leaving the software recording in the background (i.e. behind other apps or minimized) or if you’ve left the computer unattended while recording?

If you are using the glitch filter, try disabling it. Also, let us know if you used the glitch filter in 2.3.26 as well. It’s had some performance problems since we added it.

I’m capturing at 250MS/s with 2 channels (ch 1 and 2)
Recording mode is “Looping”, 3GB limit, no glitch filter.
Logic runs in background and unattended.
When I said that 2.3.26 works fine, I was meaning “while it doesn’t detect a timeout”. I’ve reduced the sampling rate to 10MS/S and the timeout occurs less frequently.
Now I’m running some system tests so I cannot perform the same tests with logic 2.3.36 until system tests are finished.

Thanks for all the details! How long do you typically leave it in the background recording?
We saw an issue with this a few releases ago and thought we fixed it. I want to run some tests with the SW recording in live mode in the background, and run it at least as long as you did.

normally I let it run for hours, depending on the test, but the performance issues start quite soon with
.36, let’s say after 5 minutes or even less.
If I zoom out let’s say to see 20 seconds of acquisition on the whole screen then the problem occurs more often (one can see that the more one zooms out, the higher the CPU load is, I guess because of the analyzers). The packets rate varies from a couple per second to > 50 per second.

I’m now performing the same test with .36 and 10MS/s (not 250MS/s) and it seems better.

Another thing I’ve noticed is that the .36 version has problems to either detect the Logic Pro 16 and connect to it or to visualize the signals. I’ve installed the .36 right now and I had to disconnect/reconnect Logic Pro 16 before the program could visualize the signals. Before, the signals were just not there, only a flat line (ground). With .26 I never had these problems.
I had similar problems a couple days ago too (Logic 16 not found).

I’ll see if I manage to find out from which version the problems occur, it would for sure help you in finding the problem.

I’ve noticed that if I load a previous aquisition where my extension is active, with .26 it is loaded much faster than with .36.

Thanks for all the details. We probably introduced the issue in 2.3.30/2.3.31. We’re investigating a related performance problem at the moment, but we didn’t expect it to affect customers as badly as you have observed.

2.3.30 was recalled after a day or two because it had a performance problem that would kill the app if it was running in the background, because the software kept computing display renders, but the display system stopped drawing them.

It seems like we may not have fixed that issue completely. We’re also working on a performance problem where the application is rendering too often, but not displaying most renders. It may make sense that this is happening more often for higher sample rates, because the internal update rate is much higher.

If we’re lucky, this is the same problem you’re seeing, in which case it will be fixed in the next release. I’ll be sure to do some side by side comparisons with 2.3.26.

We’ve just released 2.3.37 with a fix for this issue. Could you test it, and compare the performance with 2.3.26?

Hi Mark
from what I could see so far, .37 seems much better!

The visualization is fluid and I didn’t face any problems with responsiveness.
I’ll let you know if I’ll have further problems, especially what the connection to Logic 16 concerns.
Thank you very much for your support!